Classical conditioning is used to study the behavioral laws and neural substrates of learning and memory. Behavioral studies show that long- term memory for even simple forms of learning last for several months and that the strength of the memory decreases as a function of the difficulty of the task. Other experiments indicate that learning alters processing of the unconditioned stimulus as well as the conditioned stimulus and suggests that there are a number of neural pathways that are modified by classical conditioning. Electrophysiological analysis of a rabbit cerebellar slice preparation has shown that the cerebellum is modified after conditioning of the eyeblink response. Recently published experiments show that early in training, the level of Purkinje cell excitability is correlated with the level of conditioning. These changes become stronger with further training and are still in evidence one month after all training has stopped. Evidence exists for Purkinje cell potassium channel involvement in these conditioning-specific changes in excitability. Collaborative studies using positron emission tomography have examined the neural substrates of human eyelid conditioning. An initial experiment showed that there were significant conditioning-specific changes in regional cerebral blood flow (rCBF) in cerebellum and primary auditory cortex as well as posterior cingulate and prefrontal areas. These changes confirm our own animal studies as well as those of others. The changes in rCBF in primary auditory cortex were contralateral to the side of airpuff delivery and cerebellar changes were bilateral. A second experiment replicated these findings and showed that if the airpuff was delivered on the opposite side to the first experiment, the auditory cortex changes occurred on the side opposite to the airpuff. This is the first case of functional laterality reported in humans using positron emission tomography. More recent studies indicate that subjects show different patterns of rCBF when the task is made more difficult by using trace conditioning. A recently completed study of elderly subjects shows that their rCBF changes during learning are different from young subjects.